Detection of quantitative trait loci affecting haematological traits in swine via genome scanning

Background Haematological traits, which consist of mainly three components: leukocyte traits, erythrocyte traits and platelet traits, play extremely important role in animal immune function and disease resistance. But knowledge of the genetic background controlling variability of these traits is very limited, especially in swine. Results In the present study, 18 haematological traits (7 leukocyte traits, 7 erythrocyte traits and 4 platelet traits) were measured in a pig resource population consisting of 368 purebred piglets of three breeds (Landrace, Large White and Songliao Black Pig), after inoculation with the swine fever vaccine when the pigs were 21 days old. A whole-genome scan of QTL for these traits was performed using 206 microsatellite markers covering all 18 autosomes and the X chromosome. Using variance component analysis based on a linear mixed model and the false discovery rate (FDR) test, 35 QTL with FDR < 0.10 were identified: 3 for the leukocyte traits, 28 for the erythrocyte traits, and 4 for the platelet traits. Of the 35 QTL, 25 were significant at FDR < 0.05 level, including 9 significant at FDR < 0.01 level. Conclusions Very few QTL were previously identified for hematological traits of pigs and never in purebred populations. Most of the QTL detected here, in particular the QTL for the platelet traits, have not been reported before. Our results lay important foundation for identifying the causal genes underlying the hematological trait variations in pigs.


Background
In pork-producing community, various infectious diseases caused by viral or bacterial pathogens badly restrict the efficiency of pork industries, and also seriously affect human health in the whole world. To most infectious diseases, the resistance of an individual is resulted from both innate immunity and acquired immunity. The capacity of innate immunity or acquired immunity is more or less controlled by genes [1][2][3][4]. The resistance to some diseases is due to a single gene [5,6], but for most diseases it is due to multiple genes [7], among which a few genes (QTL or major genes) have relative large effects and thus play important role in disease resistance. Therefore, detection of these genes or QTL is important for disease-resistant breeding.
Haematological traits, which consist of three components: leukocyte traits, erythrocyte traits, and platelet traits, are important indicators of immune capacity of animals [8][9][10]. The influence of genetics on baseline peripheral blood cell parameters has been firmly established. In human, the heritability estimates for red blood cell, white blood cell, and platelet numbers derived from a twin study [11] were 0.42, 0.62, and 0.57, respectively, indicating a large genetic component underlies these traits. Study in baboons [12] confirmed that genetic influences account for a significant portion of the variance in these blood parameters as well as in mean corpuscular volume (MCV) and mean platelet volume (MPV). In swine, significant differences in a panel of haematological traits between Meishan and Large White pigs were observed by Clapperton et al. [1], which may have impli-cations in the resistance to infection by a broad range of pathogens and subsequent disease effects in these breeds. However, knowledge of the genetic background controlling the variability of haematological traits, is very limited, especially in swine.
With the continuously increasing of DNA molecular markers and deep development of pig genome research, a large number of QTL or major genes controlling quantitative traits in pigs have been found and located in the relevant regions of chromosomes [13][14][15]. However, compared to the production traits, QTL mapping for haematological traits has been carried out only in a few research groups. Edfors-Lilja et al. [16,17] first reported QTL on SSC1 and 8 for stress-mediated alterations in some leukocyte traits in swine, which were further confirmed later in another mapping study by the same group [18]. Reiner et al. [19] reported nine genome-wide significant and 29 putative QTL distributed on 15 chromosomes controlling baseline leukocyte traits and leukocyte traits at distinct stages of the disease model. QTL for erythroid traits were also reported by the same group [20]. 43 QTL controlling baseline erythroid traits and erythroid traits at distinct stages of the disease model were identified on 16 chromosomes, of which twelve were significant at the genome-wide level. In a very recent study [21], a total of 101 QTL, including 46 genome-wide significant QTL, regulating baseline erythroid traits at different age stages were found on all pig chromosomes except for SSC15 and SSC18. For platelet traits, there are no reports of QTL mapping studies in swine up to now.
The immune response to "stress" is important indicator of disease resistance capacity of animals. In the present study, we report the results of QTL mapping for the stress induced immune response of pigs which were challenged with swine fever vaccine. Unlike the previous studies mentioned above, where F2 animals derived from cross of two breeds was used as mapping population, we used purebred animals with half-and full-sib families as mapping population. 18 haematological traits, including seven leukocyte traits, seven erythrocyte traits, and four platelet traits, were measured after they were inoculated with the swine fever vaccine.

Phenotypes
The means and standard errors of the phenotypic values of the 18 haematological traits in the mapping population are presented in Table 1. These results are generally consistent with previous results [19][20][21]. Significant differences (P < 0.05) between measurements before (on day 20) and after (on day 35) inoculation with the swine fever vaccine were observed for all traits except GR%, LY% and MO%. Differences between the three breeds are significant for most of the leukocyte traits and two of the four platelet traits, but not significant for all erythrocyte traits except MCHC. No significant differences between sexes were observed for all traits except RBC and MCHC.

QTL for the 18 haematological traits
The QTL detected with FDR < 0.10 for the 18 haematological traits are presented in Table 2. For the leukocyte traits, only three QTL, including one with FDR < 0.05, were detected for GR, MO% and WBC, respectively. For the erythrocyte traits, 28 QTL, including 21 with FDR < 0.05, were detected: ten for MCHC, seven for RDW, five for MCV, three for MCH, two for HGB, and one for HCT. For the platelet traits, four QTL, including three with FDR < 0.05, were detected: two for PCT and two for PDW. The positions of all of the detected QTL are shown in Fig. 1.

Discussion
So far, there are only limited reports on QTL mapping for haematological traits in pigs [16][17][18][19][20][21]. All these studies used F2 animals as mapping population and only one category of haematological traits (leukocyte traits or erythrocyte traits) were involved. In the present study, we used purebred animals as mapping population and all three categories of haematological traits were analyzed. There is large inconsistency between the QTL for leukocyte and erythrocyte traits reported in the present and the previous studies. For the leukocyte traits, no QTL are in agreement with those reported in the previous studies. In particular, for most traits, the QTL found here and in the previous studies were mapped to different chromosomes. For example, for the trait WBC, we mapped two significant QTL on SSCX, but the QTL for this trait were found on SSC1, SSC8, SSC10, and SSC12 in the previous studies [13][14][15][16]. For the erythrocyte traits, only a few QTL are in agreement with those reported in their studies. For HGB and MCHC, we mapped a QTL at 31 cM on SSC2 and a QTL at 19 cM on SSC16, respectively, which overlap those reported by Reiner et al. [20]. For RDW, we mapped a QTL at 20 cM on SSC1, which overlaps that reported by Zou et al. [21]. Large inconsistency also exists between the results from Reiner et al. [20] and those from Zou et al. [21]. Both studies involved in mapping QTL for erythrocyte traits in swine and employed the regression approach and a F2 design. 43 and 101 QTL were found by Reiner et al. [20] and Zou et al. [21], respectively. However, only four are in accordance with each other. A major possible reason for these inconsistencies is that the measurements for the traits are different in different studies. In the study of Edfors-Lilja et al. [17], the leukocyte traits were measured immediately before and after the pigs were transported from the farrowing pen to a finishing unit and mixed with non-littermates when they were about three months old. In the study of Zou et al. [21], the    [19,20], the leukocyte and erythrocyte traits were measured on the day (when the pigs were approximately 100 days old) of and shortly before infection with 50 000 sporocysts of Sarcocytis miescheriana of and on days 14, 28, and 34 after infection. While in the present study, all haematological traits were measured one day before and 14 days after the inoculation with the swine fever vaccine when the pigs were 21 days old. The immune system of pigs is very sensitive to different external stresses. The genetic mechanism of immune responses to different external stress and in different physiological stage is very complex and currently unknown. Therefore, the phenotypes for the same haematological parameters measured in different studies may represent different traits, although they may be highly correlated. Another possible important reason for the discrepancies is that different mapping populations were used in different studies, in which the QTL segregation status may be different. In all cited studies, QTL were detected in crossbred populations while they were detected in purebred populations in the present study. QTL mapping for platelet traits in swine has not been reported yet. However, QTL for platelet traits had been studied in mice [22,23]. Among the QTL found in the two studies, a QTL for PLT was mapped to 52 Mb [22] and 51 Mb [23] on MMU7, respectively. This region is syntenic with the region on SSC2, where we found a putative QTL (P < 0.05, data not shown) for PLT in swine. In addition, Chueng et al. [23] mapped a QTL for PLT in mice to 48 Mb on MMU17, which is syntenic with the region of 71 cM on SSC7, where we also found a putative QTL (P < 0.05, data not shown) for PLT in swine.
Similar to the results of previous studies [20,21], some QTL located at the same or nearby positions were found to have pleiotropic effects on two to three traits in the present study. These include the QTL located at 38 cM on SSC2 (affecting PDW and PCT), the QTL located at 124 cM on SSC9 (affecting MCH and MCHC), the QTL located at 8 and 9 cM on SSC12 (affecting HGB and HCT), and the QTL located at 51 cM on SSC14 (affecting MCH, MCHC, and RDW). These pleiotropic QTL contribute largely to the correlations between traits. For example, we can imagine a high genetic correlation between traits MCH and MCHC, and indeed, two QTL, one at 124 cM on SSC9 and the other at 51 cM on SSC14, were found having significant effects on both traits.
The candidate genes related to leucocyte traits have been discussed in detail by Wattrang et al. [18]. However, none of these genes are located in the QTL regions identified in this study. For the erythrocyte traits, three candidate genes were reported, including the porcine stem cell growth factor receptor (KIT) gene [24], the erythropoietin (EPO) gene [25], and the erythropoietin receptor (EPOR) gene [26]. The KIT gene is located at p12-p21on SSC8 and is essential for normal hematopoiesis, melanogenesis, and germ cell development [27]. The EPO gene is located at p16-p15 on SSC3 and plays important role in controlling the level of circulating erythrocyte mass by promoting erythroid differentiation and initiating haemoglobin synthesis [28]. The EPOR gene is located at q1.2-q2.1 on SSC2 and is critical for erythroid proliferation and differentiation [26]. In the present study, no QTL was found around the position of the KIT gene, a putative QTL (P < 0.01, data not shown) affecting HGB was mapped between markers SW256 and S0141 on SSC2 which is close to that of the EPOR gene, and a QTL affecting MCV was mapped between markers SW2618 and SW836 on SSC3 which is near the position of the EPO gene.

Conclusions
Very few QTL were previously identified for hematological traits of pigs and never in purebred populations. In this study, 35 QTL with FDR < 0.10 were identified for 18 haematological traits: 3 for the leukocyte traits, 28 for the erythrocyte traits, and 4 for the platelet traits. Most of the QTL detected here, in particular the QTL for the platelet traits, have not been reported before. Our results lay important foundation for identifying the causal genes underlying the hematological trait variations in pigs.

Measurement of phenotypic traits
The piglets were inoculated with the swine fever vaccine on day 21 after birth. Blood samples were collected from each piglet one day before the inoculation (day 20) and two weeks after the inoculation (day 35), respectively. The samples were directly injected into eppendorf tubes containing 60 μl of 20% EDTA in phosphate-buffered saline (PBS). 18 hematologic traits (Table 1) were measured immediately after sample collection using a TEK-II mini automatic hemocyte analysor (Jiangxi Tekang Science and Technology Company, China) with swine specific parameter configuration at Beijing Xiyuan Hospital, China.

Genetic markers
Genomic DNA was extracted from ear tissues of all piglets and their parents (445 animals in total) using a standard phenol/chloroform method. Microsatellite markers were initially selected from the USDA-MARC porcine reference map http://www.marc.usda.gov/ and were then analyzed for their polymorphisms using DNA from the boars. A final set of 206 informative markers distributed on 19 chromosomes (18 autosomes and chromosome X) was genotyped for all individuals. The total length covered by these markers is 2261.7 centi-Morgans (cM) with an average distance of 11.6 cM between adjacent markers according to the MARC map. The number of markers and their mean polymorphic information content (PIC) on each chromosome are shown in Table 3. For the QTL analysis, the linkage maps of the selected markers were derived directly from the MARC map.

QTL analysis
A QTL interval mapping analysis was performed using the variance component approach [29][30][31]  It is generally regarded that the LR follows approximately a χ 2 distribution with degrees of freedom between one and two [29]. To be more conservative, we assumed here it follows a χ 2 distribution with two degrees of freedom. In addition, in order to avoid the increase of false positive rate caused by multiple tests, the FDR (false discovery rate) control approach [32,33] was adopted to determine the significance thresholds based on the P values from the χ 2 distribution. The positions with FDR < 0.05 were considered harboring significant QTL. Let m be the total number of tests involved in the analysis and P 1 ≤ P 2 ≤<≤ P m be the ordered observed P-values for the m tests, the FDR for P i is calculated as Since the QTL analysis was performed at 1 cM intervals along each chromosome, each genome scan is equivalent to 2,261 tests given the assumed genome length. As these tests are not independent, Weller et al [34] recommended as a cautious choice considering the number of markers genotyped. This number was 206 in the present study, The LOD-drop off method [34] was used to estimate the confidence intervals of the QTL positions.
Since the variance component method requires normal distribution of the phenotypes, we tested the normality of all traits. It turned out that most of the traits were not normally distributed. The Box-Cox transformation was performed for the traits which severely depart from normal distribution before the QTL analysis was carried out.
Mapping quantitative trait Loci for milk production and health of dairy cattle in a large outbred pedigree. Genetics 1998, 149 (4)